Tyler Cowen — Hayek, Keynes, & Smith on AI, animal spirits, anarchy, & growth

The conversation opens with Cowen's book about the greatest economists and Keynes's portrait of the 'master economist' as mathematician, historian, statesman, and philosopher. Dwarkesh jokes that the description could fit Cowen; Cowen rejects the comparison, especially the breadth of Keynes's public achievements. This framing establishes a theme of the discussion: how to judge an economist not only by a few celebrated ideas but by the breadth, consistency, and limitations of their work.

Dwarkesh uses Keynes's account of investment to ask whether much historical investment was irrational: investors may be driven by excitement and the satisfaction of building, even when returns disappoint. Cowen traces the idea to Adam Smith's warning that entrepreneurs are often overoptimistic. He says private investment returns are highly skewed: established venture firms can earn a great deal, while many small businesses make little or nothing. Cowen accepts Keynes's intuition but cautions that Keynes lacked modern distributional data, and that the commonly repeated claim that innovators capture only 2% of social gains is not firmly measured. The examples of Picasso and spillover influence illustrate how hard that measurement is.

Keynes also argued that genuine long-term investing is difficult compared with speculating on how other investors will behave. Cowen says Keynes believed short-run speculators could beat markets and may have done so at times, though luck and special insight are hard to distinguish. He thinks Keynes's larger concern was that short-term profit seeking could undermine long-term investment, which Keynes wanted society to direct. Cowen does not dismiss the argument, but says it may be historically contingent: investors in Keynes's era had less data, weaker international supply chains, and greater geopolitical disruption. The exchange also notes that historical market efficiency is difficult to test because trading and price data are incomplete.

The discussion of passive investment raises fears that a small number of asset managers could weaken corporate oversight or tacitly reduce competition. Cowen is not very worried so long as marginal actors can still trade on information; he says research activity has not weakened, though few public companies could be a problem. A more serious risk, in his view, is coordination through clearinghouses and payment networks: if participants are excluded from the system for refusing to cooperate, the infrastructure can enforce collusion. He also suggests that trading may be privately excessive but socially useful because it spreads information, and that human overconfidence can sometimes support that activity.

On the size of finance, Cowen argues that its share of wealth is more informative than its share of GDP. He estimates US finance has remained near 2% of wealth for decades, which he sees as less alarming than an 8% GDP share suggests; as wealth grows relative to output, finance's GDP share may rise without being sinister. The exchange illustrates his broader preference for choosing the denominator and comparison carefully before inferring that an industry is too large. It closes the segment by contrasting Keynes's idea that people are excessively risk-seeking with the more familiar claim that people are risk-averse, leaving the reconciliation for the next part.